Over 10 weeks, our team delivered strategic product recommendations by prototyping features, designing high-fidelity Figma wireframes, and iterating through 13+ usability and A/B tests. We collaborated to translate insights from 40+ interviews, 100+ surveys, and competitive analysis into actionable design directions.
Through this project, we designed features informed by key student pain points around AI usage, trust, and context-switching. We mapped academic workflows, identifying four core user personas, insights that will shape Perplexity’s strategic approach to building for higher education.
We proposed a Citation Generator integrated into Perplexity’s Research mode, which supports multiple citations at once, customizable formats (APA, MLA, Chicago), and allows one-click copy/paste of a structured bibliography.
This feature is designed to minimize context-switching and promote answer and source credibility, a reoccurring value students expressed in our surveys and interviews.
GATHERING THE DATA
We wanted to better understand what types of AI tools students most frequently use and how they interact with AI across their academic workflows.
Our research revealed that credibility is the single biggest motivating factor for students adopting AI tools in their academic work. While many appreciated Perplexity’s linked citations, they wanted deeper transparency and easier ways to verify and manage sources across their workflows.
During user testing, an unexpected insight emerged—several students independently suggested that a sidebar citation generator might better fit their workflow than the existing modal popup design. This sidebar idea sparked curiosity, and I quickly created a second prototype to conduct testing with.
MODAL POP-UP
Most testers initially expressed no issues with the modal, but many were interested in a sidebar version.
SIDEBAR
Students preferred a sidebar citation generator over a popup for its ability to keep all sources visible and preserve context during citation.
Through 13 additional interviews, we narrowed down our designs into features students expressed interest in using and found value in. Many students expressed interest in the Citation Generator, a built-in citation generator with customizable styles located within the “Sources” tab of a search result.
In user testing interviews, students questioned what it means for a source to be credible, and student researchers shared that they are most interested in a source’s number of citations and whether it’s been peer reviewed.
Students were confused about the difference between the citation confirmation stage and the final citation screen, so I changed subheadings to clearly show what each line was referencing, adding number of citations and peer review status as well.
BEFORE
AFTER
Added number of citations, peer review status, and “Source Title” before article name to differentiate between the source and actual citation.
Perplexity is an emerging AI search platform focused on delivering fast, reliable answers with integrated citations. Students use it for research, homework, and exploring new topics, but workflows are often fragmented and trust in AI-generated sources remains a challenge.
The Product Space team at UCLA was asked to identify product opportunities that would make Perplexity easier to use, more relevant for academic tasks, and ultimately more trusted by students as a research and learning tool.
EARLY CITATION GENERATOR DESIGN
Most testers initially expressed no issues with a modal, but many were interested in a sidebar version, prompting further exploration.
FINAL ITERATION
Students preferred a sidebar citation generator over a popup for its ability to keep all sources visible and preserve context during citation.
We created four personas—the Scholar, Inquirer, Creator, and Professional—based on student quotes and behavioral patterns around AI use. By clustering their needs and frustrations into themes like accuracy, input capabilities, and research workflows, we identified key opportunity areas to inform feature ideation.
With our synthesized insights, we began ideating and designing features would cater to students’ varying needs across learning, researching, and content creation and enhance their Perplexity user experience.
REFLECTIONS
Lessons Learned
This project gave me the opportunity to dive deeper into AI design while exploring a more open-ended space. I really enjoyed working with a team again, learning from each other, and building on our strengths across research and design.
One of the biggest lessons was how often our assumptions about users didn’t match reality. We came in expecting people to use AI in certain ways, but through interviews and testing, we uncovered use cases that completely shifted our direction. Features like the moodboard, which we were excited about early on, ended up being too niche. That insight helped us refocus on solving more relevant problems like citation generation.
The fast pace challenged me to synthesize insights quickly, test early, and stay focused on our users. I also learned the valuable lesson of designing with users, not just for them!
Feedback is a gift and a learning opportunity!
Some “failed” concepts surfaced surprising insights about student AI usage, patterns that informed stronger iterations and helped the team align on what to prioritize
Iterate fast when research surprises you
Some of our biggest breakthroughs came after usability sessions that challenged our assumptions! Instead of sticking to the original flow, I quickly reworked prototypes, tested again the next day, and repeated the cycle. This reinforced one of my favorite quotes: "Great products aren't derailed by twists and turns; they are defined by them," in Practice by Figma!
Even while focused on MVP features, I learned to document potential integrations and long-term opportunities, like how future AI model updates or collaboration tools like Notion or Canvas could fit into our current flow. Keeping those “next steps” in mind early made our designs more flexible and sharpened our product thinking!
Thank you so much for reading!
